Mapping the environmental sensitivity to oil spill and land use/land cover using spectrally transformed Landsat 7 ETM data

Abstract

Accidents involving oil spill in the marine environment cause a number of impacts on biological communities and impose severe damages on human-use resources. Environmental sensitivity index (ESI) maps are an integral component of oil spill contingency planning and emergency response. ESI maps can also be used for environment management planning by adding land use/land cover (LULC) information on shoreline classification, biological and human-use resources. We have assessed the adequacy of restored ETM+/Landsat 7 images to generate ESI and LULC maps for a stretch of coastline between the cities of Areia Branca (RN) and Fortim (CE) in the Potiguar sedimentary basin, Northeast Brazil. This has been done by comparing results from segmentation by region growth and unsupervised classification of images generated by standard and selective principal components, IHS transform, soil adjusted vegetation index, and linear spectral mixture model. The assignment of image classes to map themes was aided by visual interpretation of digitally processed images and ground truth. Results clearly indicate that image restoration followed by fusion of multispectral and panchromatic ETM images via IHS transformation is the most versatile image product to be segmented and classified. The mapping approach is based on information from both image texture and statistics, and has the advantage of reducing interpretation bias introduced by analyst during contextual editing. We also show that the hybrid mapping scheme that uses automated and visual image interpretation is particularly useful for mapping the ESI and LULC.